agricultural landscape value and irrigation water policy

Agricultural Landscape Value andIrrigation Water Policy

Mara Thiene and Yacov Tsur1

(Original submitted February 2012, revision received February 2012, accepted Feb-ruary 2012.)

Abstract

Water is a limiting factor of agricultural production in an increasing number ofregions. There is also ample empirical evidence to suggest that the economicvalue of agricultural landscape is substantial, which has been used to justify agri-cultural support programs in developed economies. We investigate the linkbetween the environmental amenity of agricultural landscape and the value ofwater in crop production. We find that the environmental externality gives rise toa social derived demand for water which differs from the market-based (private)derived demand for water. Policy implications regarding irrigation water alloca-tion and pricing are drawn. An empirical example illustrates the methodologyand main findings.

Keywords: Environmental amenity; hedonic pricing; social value of water; willing-ness-to-pay.

JEL classifications: Q15, Q51, Q58.

1. Introduction

Land and water are primary inputs in crop production and, as such, in the ensuing(by-product) supply of environmental amenities such as farm landscape and eco-system services.2 These environmental amenities have often been used to motivateagricultural policies such as land set-aside payments and land targeting or zoning, yettheir implications regarding water allocation policies remain unexplored. The latter isparticularly relevant for perennial crops, where a water shortage below the minimal

1Mara Thiene is an Associate Professor in the Department of Land, Environment, Agricultureand Forestry, University of Padua, Viale dell’Universita’ 16, 35020 Legnaro (Pd), Italy. E-mail:[email protected] for correspondence. Yacov Tsur is a Professor in the Departmentof Agricultural Economics and Management, The Hebrew University of Jerusalem and TheCenter for Agricultural Economic Research, POB 12, Rehovot 76100, Israel. E-mail:[email protected] often entails negative externalities (pollution run-off due to fertilisers and pesticides,emission of greenhouse gases associated with tillage practices), the regulation of which has beenstudied extensively. Here, we focus on the positive externality associated with agricultural

landscape.

Journal of Agricultural Economics, Vol. 64, No. 3, 2013, 641–653doi: 10.1111/1477-9552.12016

� 2013 The Agricultural Economics Society. Published by Blackwell Publishing,

9600 Garsington Road, Oxford OX4 2DQ, UK, and 350 Main Street, Malden, MA 02148, USA.

water needed for the crop survival inflicts a large loss, thus deterring farmers fromgrowing the crop. In this work, we study implications of farm landscape vis-�a-vis irri-gation water policies.

Our work is related to the growing literature on the external effects of agriculturallandscape in the context of urban–rural land allocation (see, e.g. McConnell, 1989;Lopez et al., 1994; Brunstad et al., 1999; OECD, 2000; Peterson et al., 2002;Cavailh�es et al., 2008; Morey et al., 2008; Bergstrom and Ready, 2009). Farmlandamenities and disamenities have been estimated in a number of countries (see the sur-veys by Cavailh�es et al., 2008; Bergstrom and Ready, 2009), including Austria (Hackland Pruckner, 1997), USA (Halstead, 1984; Bergstrom et al., 1985; Beasley et al.,1986; Ready et al., 1997; Irwin and Bockstael, 2002), UK (Hanley et al., 1998; Scarpaet al., 2007), Canada (Bowker and Didychuk, 1994) and Israel (Fleischer and Tsur,2000, 2009). Agricultural landscape values have been differentiated according to basicactivities (tilled land, pasture and woodland) by Drake (1992), Brunstad et al., (1999)Schl€apfer and Hanley (2003), and Fleischer and Tsur (2009) provided landscapevalues for a variety of crops. Policy implications of agricultural landscape values havefocused mainly on land policies, such as land subsidies, targeting and zoning.

As the demand for water by other sectors (mainly households) increases withpopulation growth and higher living standards, water is becoming the limiting factorof agricultural production in many regions. The result is a transfer of cultivated areainto fallow or urban areas – a process that cannot easily be reversed. It is thereforeimportant to understand the links between farmland policies and the demand for irri-gation water. Here, we study such links in the context of policies aimed at internalis-ing crop-specific landscape values and show how such policies can be implemented viairrigation water policies.

We first provide a theoretical framework aiming at incorporating the landscapevalue within the value generated by irrigation water. We extend the traditional notionof private derived demand for water to define the social derived demand and showhow the landscape values can be internalised via water allocation policies. Then, wemake use of an empirical case study based on the hedonic price approach to illustratethe application of the methodology. In particular, based on actual property transac-tions data collected in the north-east of Italy, we estimate farm landscape values ofagricultural crops and discuss potential effects on policies related to the allocation ofirrigation water.

The article proceeds as follows: section 2 summarises the notion of agriculturallandscape value. Section 3 focuses on the social derived demand for water and its allo-cation policies. Section 4 provides an empirical example and section 5 concludes.

2. The Landscape Value of Farmland

We consider a region whose land area is allocated between urban (dwelling, services,industry) and farming. Farmland is allocated among K crop groups (open space andparks could be considered as a crop group), scattered spatially across the region. LetLi ¼ ðl1i; l2i; . . .; lKiÞ represent the allocation of farmland to the K crops in the localityof household i, defined as the surrounding area affecting the household’s utility. Occa-sionally it will be useful to represent Li in the equivalent form ðlki;L�kiÞ, whereL�ki ¼ ðl1i; . . .; lk�1i; lkþ1i; . . .; lKiÞ represents land allocation to all crops excluding thekth crop.

� 2013 The Agricultural Economics Society.

642 Mara Thiene and Yacov Tsur

Household i derives utility from a composite market good fi with a unit price andfrom dwelling characteristics ðZi;LiÞ according to the utility function uiðfi;Zi;LiÞ,where Zi represent house-specific characteristics such as a property’s area, number ofbedrooms, number of bathrooms, age, type (townhouse, apartment, villa). The indi-rect utility is defined by:

viðyi;LiÞ ¼ maxfi;Zi

uiðfi;Zi;LiÞ s.t. fi þ hðZi;LiÞ� yi; ð1Þ

where h (Z,L) is the (equilibrium) housing hedonic price function (Rosen, 1974) andyi is the household’s income.

Given L�ki, the willingness-to-pay to preserve lki (crop k area in household i’s local-ity) against the alternative of not cultivating lki (leaving it fallow) is denoted WTPki

and defined by:

viðyi; 0;L�kiÞ ¼ viðyi �WTPik; lik;L�kiÞ: ð2ÞThe above WTP index is defined with respect to a particular alternative, namely

that of leaving the lik area fallow. Other alternatives, for example, replacing the likarea with another crop or transforming lik into an urban area, will give rise to differentWTP indexes.

Summing over all households in the region gives the total willingness-to-pay (WTP)to preserve crop k area (against the alternative of leaving it fallow):

WTPk ¼Xi

WTPik: ð3Þ

Notice that any hectare of crop k appears in the WTP of all households that resideclose enough to this hectare (i.e. all households for which this hectare is in their vicin-ity). In the above summation, therefore, the landscape value of any of crop k’shectares appears in many WTPki terms, which is a manifestation of the public-goodnature of agricultural landscape.

Let Ak represent crop k’s area (number of hectares) in the region. The per-hectarelandscape value of crop k is:

Wk ¼ WTPk

Ak: ð4Þ

3. Landscape Values and Water Allocation

The derived demand for irrigation water is the value marginal product of water gener-ated by marketable output (crop yield). We extend this notion to include landscapeamenity. Incorporating landscape values gives rise to a distinction between the privateand social demand for water. We first explain this distinction and then discuss impli-cations for water allocation.

3.1. Private vs. social demand for irrigation water

Crop k’s output is produced by the input vector ðQk;Xk;AkÞ according to the produc-tion function FkðQk;Xk;AkÞ, where Qk is water input, Ak is land input and Xk is a vec-tor of all other inputs (e.g. labour, capital, fertilisers). Assuming that Fk admitsconstant returns to scale, it can be expressed as FkðQk;Xk;AkÞ ¼ Ak

~fkðqk; xxÞ, where


Landscape Values, Hedonic Price, Water Policy 643

qk ¼ Qq=Ak is the per-hectare water input, xk ¼ Xk=Ak is the vector of per-hectareinputs other than water and land and ~fk is the per-hectare production function.

Let

~x�kðqkÞ ¼ arg maxxk pk~fkðqk; xkÞ � px � xkn o

ð5Þwith the ensuing profit

~pkðqkÞ � pk~fkðqk; ~x�kðqkÞÞ � px � ~x�kðqkÞ; ð6Þwhere the output and input prices (pk and px) are suppressed as arguments of ~pkð�Þ forconvenience. Under common conditions, involving input–output prices and produc-tion technology properties (increasing and strictly concave over an admissible inputrange), ~pkð�Þ is positive above some critical water input q

k[ 0. The x-inputs demands

are given by:

x�kðqkÞ ¼ ~x�kðqkÞ if qk � qk

0 otherwise

�; ð7Þ

and the water-yield function is defined by:

fkðqkÞ ¼ ~fkðqk; x�kðqkÞÞ if qk � qk

0 otherwise

�: ð8Þ

We maintain the common properties that fkð�Þ is increasing and strictly concaveover ½q

k; �qk�, 0\ q

k\ �qk � 1, where �qk is the water input at which fkð�Þ reaches its

maximum. The water demand function

~ukðpqÞ ¼ arg maxqk pkfkðqkÞ � pqqk � pxx�ðqkÞ

� � ¼ f 0�1k ðpq=pkÞ ð9Þ

is thus well defined over the water price range ½0; �pkq�, where�pkq � pkf

0kðqkÞ: ð10Þ

(To verify the rightmost term of (9) note, recalling (5), that the term multiplyingx�0ðqÞ vanishes.)

When the water price is pq 2 ½0; �pkq�, crop k’s growers will demand the per-hectarewater input ~ukðpqÞ if the ensuing profit pkfkð~ukðpqÞÞ � pq~ukðpqÞ is non-negative; other-wise, they will demand no water. Define gk as the break-even water price, satisfying:

pkfkð~ukðgkÞÞ � gk~ukðgkÞ ¼ 0: ð11ÞThe per-hectare water demand by crop k’s growers can now be defined by:

ukðpqÞ ¼~ukðpqÞ if pq 2 ½0; gk�0 if pq [ gk

�: ð12Þ

Figure 1 provides a graphical illustration. At a water price pq, crop k growers willdemand the irrigation water input ~uðpqÞ ¼ f 0�1ðpq=pkÞ if the revenue (the areabeneath pkf

0ð�Þk between q

kand ~uðpqÞ) does not fall short of the water cost pq ~uðpqÞ. At

a water price gk, the revenue just equals the water cost (in Figure 1, the areagkb~uðgkÞ0 just equals the area q

kab~uðgkÞ). When the water price pq is above gk, any

use of a positive quantity of irrigation water will generate a loss, hence no irrigationwater will be demanded. At a water price below gk, water will be demanded up to thepoint where its value of marginal product just equals the water price, that is, along thepkf

0kð�Þ curve, as depicted in Figure 1.



We refer to ukð�Þ as the private water demand per hectare (Figure 1). The ‘private’qualifier signifies that this demand is driven solely by market forces and ignores extra-market effects such as the landscape value of farmland.

Growers are compensated for the output they sell in the marketplace, but not forthe landscape amenities they generate. The former is given by the producer surpluspkfkðuðpqÞÞ � pquðpqÞ and the latter by Wk, specified in equation (4) – both indexesare measured in € per hectare. The private demand for irrigation water, discussedabove, accounts only for surpluses associated with the harvested crop. The socialdemand, defined below, incorporates the landscape value generated by the waterinput.

Let

wk ¼ Wk

qk

ð13Þ

represent the surplus per m3 generated by the water input qkdue to the landscape

amenity.3 Suppose that wk [ gk, as depicted in Figure 1. We know from equation(12) that when the price of water pq exceeds gk, farmers will cease the production ofcrop k (after some time, if not immediately). However, from a social point of view, aslong as pq � wk, it pays to use at least the water input q

k, as this water input generates

the surplus wk qk(= the landscape value Wk), which exceeds the water cost

pq qk. Thus, as long as pq � wk, the demand for water owing to the landscape value

is qk. Combining the latter with the private demand, specified in (12), gives the social

demand for water when wk [ gk:

wk

m3 per hectare (q)

Euro per m3

(pq)

kq

Private (derived) demand for water

Social (derived) demand for water

gk

m

m

0)/(')(~ 1kkkk pgfg

)()(~ '1 qfpq kkk

kqp a

bc

Figure 1. Social and private demands for water

3As Wk is measured in € per hectare and qkis measured in m3 per hectare, wk is measured in €

per m3 units.



uskðpqÞ ¼

ukðpqÞ if pq 2 ½0; gk�qk

if pk 2 ðgk;wk�0 if pq [wk

8<: : ð14Þ

(If wk � gk, the social and private demands are the same.) The social demand forwater is depicted together with the private demand in Figure 1.

3.2. Optimal water allocation

The optimal water allocation is obtained at the point where the social derived demandfor water and the marginal cost of water supply intersect.4 Figure 1 depicts the socialand private demands for water. Suppose, for simplicity, that the cost of water supplyis linear in the quantity of water supplied (measured, say, in million m3 per year), sothat the marginal and average costs of water supply are equal to the same constant,denoted m. The marginal cost pricing rule implies setting pq ¼ m (pq indicates theprice of water). If pq ¼ m\ gk, then crop k growers will demand the quantityuðpqÞ ¼ usðpqÞ, so the marginal cost pricing rule gives rise to the (socially) efficientallocation without any intervention. Arguably, there is a case for compensating cropk growers for the landscape value of Wk € per hectare (or w € per m3) they provide,but efficiency is achieved irrespective of such compensations.

If, however, pq ¼ m 2 ðgk;wkÞ, then the private demand for water vanishes,uðpqÞ ¼ 0, while the social demand equals usðpqÞ ¼ q

k(see Figure 1). In this case,

some intervention is needed to implement the socially optimal allocation. Such anintervention can take different forms. One possibility is to pay crop k’s growers theper-hectare subsidy Wk ¼ wk q

k. Another possibility is to subsidise the first q

km3

of irrigation water by the rate (m � g) € per m3 for each hectare of crop k produced.We further discuss these policies in the concluding section, with the hindsight of theempirical example considered next.

4. An Empirical Illustration

We estimate farm landscape values in Vicenza, Italy, for vineyards and orchards anddiscuss their possible effects on irrigation water allocation and pricing.

4.1. Estimating landscape values

There are different ways to estimate agricultural landscape values (see Fleischer andTsur, 2009, for estimates based on contingent valuation as well as a discussion ofother examples). Here, we use the hedonic price method based on actual propertytransactions data collected in the Vicenza region of north-east Italy. This metho-dology has been widely used in valuing extra-market traits in general and environmen-tal amenities in particular (see Palmquist, 2005; Bockstael and McConnell, 2007, andreferences they cite).

4 To simplify and allow a sharp focus on the effects of landscape values, we ignore long-termconsideration associated with fixed (capital) cost of water supply (for a discussion of the effects

of fixed cost on water pricing rules see Tsur, 2009).



Underlying a hedonic price valuation is the hedonic price function:

P ¼ hðZ;LÞ;characterising equilibrium in the region’s housing market, where P is the price of aproperty with specific characteristics Z (number of rooms, location, garden area, etc.)and environmental characteristics (surrounding cropland in the present context) L.On the demand side, household i derives utility from housing characteristics Zi, thesurrounding cropland allocation Li and consumption of a composite good ni. In equi-librium, the household’s chosen location Li, housing characteristics Zi and compositegood ni, maximise utility subject to the budget constraint hðZi;LiÞ þ ni ¼ yi. Achange in Li may lead to a change in any of the choice variables and may also affectthe supply of housing characteristics, which in turn can lead to a change in the equilib-rium hedonic price function h(�). This feature complicates welfare evaluation and mayturn the estimation of hedonic price functions intractable (Ekeland et al., 2004).

If, however, the transaction costs of moving are such that the change in Li does notinduce relocation, then the ensuing welfare change can be deduced from the ensuingchange in the average property value. In such a case, a change in the cropland alloca-tion from the original situation ðlki;L�kiÞ to ð0;L�kiÞ induces the change:

DPik ¼ hðZi; lki;L�kiÞ � hðZi; 0;L�kiÞ ð15Þin the property value (obviously, DPik ¼ 0 if lik ¼ 0 in the original allocation).

We adopt the translog specification for the hedonic price function:

pi ¼Xnzm¼1

zimam þXKk¼1

‘kibk þXk

Xj

‘ki‘jickj ð16Þ

with ckj ¼ cjk, where lower-case variables represent the log of their respective upper-case variables for continuous variables (for dummy, 0-1, variables the lower-case andupper-case variables are the same). A positive landscape amenity of crop k entails apositive bk, whereas diminishing marginal effects imply that the ckk are negative. Thesign of the cross effects ckj; k 6¼ j, depends on whether crops k and j landscapes arecomplementary (positive cjk) or substitutable (negative cjk).

The cropland data used in the estimation below consist of dummy variables, indi-cating the presence of the k’s cropland near the property, that is, ‘ki ¼ 1 if crop k landexists within 1 km of household i’s dwelling and ‘ki ¼ 0 otherwise.

Thus, under the translog specification (16), equation (15) becomes:

DPki ¼ Pi 1� e�bk�

Pjckj

h iif ‘ki ¼ 1

0 if ‘ki ¼ 0

(; ð17Þ

where Pi ¼ epi is the actual property price. This change in property value measuresthe present value of an indefinite benefit stream. The annual equivalent value isobtained by multiplying DPki by the interest rate r. The annual WTP to preserve cropk’s landscape in subregion (municipality, district, county) M, defined in (3), specialiseshere to:

WTPMk ¼

Xi2M

rDPki: ð18Þ

Dividing WTPMk by the number of hectares of crop k in subregion M (denoted AM

k )gives WM

k – the per-hectare WTP for crop k, specified in (4). Dividing WMk by qM

k(the



minimal irrigation requirement of crop k in subregion M) gives wMk – the WTP per

cubic metre allocated to crop k in subregion M, specified in (13).

4.2. Data

Data pertaining to 121 housing transactions, carried out during 2007 in the Vicenzaregion, were collected from real-estate agents that carried out the transactions andcould identify the property and its characteristics, including the agricultural landscapeand livestock farming within its vicinity, where orchard, vineyard and meadows arewithin the household’s vicinity if they exist in the surrounding area within 1 km fromthe household, and livestock are within the household vicinity if a livestock farm canbe seen from the household (which may be further away than 1 km).5 The specificdwelling characteristics include dwelling type (apartment, house or villa), dwellingarea (m2), garden area (m2), number of bedrooms, number of bathrooms, mainte-nance level (grades 1–3), age of dwelling (years), distance to train station (minutes ofdriving), proximity to industrial or commercial areas (grades 0–2) and municipality(12 dummy variables for the 13 municipalities).

Four crop groups are considered: vineyards, orchards, meadows and livestock.These crops comprise the majority of the rural landscape of the area under study,where livestock refers to farms with barnyards. This activity entails a negative exter-nality due to noise and smell, but it may also give rise to a positive externality if cowsare allowed to graze in the meadows. The net impact of livestock consists of the sumof these conflicting effects. Due to the hilly character of the area of study, field cropscomprise a negligible fraction of land and are therefore not included.

Table 1

Descriptive statistics.

Variable Description Mean Max Min SD

Price Transaction values (€) 271,467 1,150,000 78,000 163,863

Dwelling area Area of the property (sqm) 151.9 370.0 40.0 73.5Garden area Area of the garden (sqm) 299.7 4000.0 0.0 654.3Bedrooms Number 2.7 5.0 1.0 0.8

Bathrooms Number 2.0 4.0 1.0 0.8Maintenance 1 = good, 0 = not good 0.5 1.0 0.0 0.5Age of house Number of years 22.4 307.0 0.0 32.7

Villa =1 if villa, 0 = no 0.0 1.0 0.0 0.2Commercial =1 if commercial area nearby, 0 = no 0.1 1.0 0.0 0.3Train Driving time to train station (minutes) 20.7 35.0 5.0 6.8Cattle =1 if cattle nearby, 0 = no 0.1 1.0 0.0 0.3

Orchards =1 if orchards nearby, 0 = no 0.3 1.0 0.0 0.5Vineyards =1 if vineyards nearby, 0 = no 0.6 1.0 0.0 0.5Meadows =1 if meadows nearby, 0 = no 0.6 1.0 0.0 0.5

Urban parks =1 if urban parks nearby, 0 = no 0.6 1.0 0.0 0.5

Note: The qualifier ‘nearby’ means ‘within 1 km of the property’.

5The landscape variables are dummies, assuming the value 1 if the landscape (or livestock) exists

within the household’s vicinity, and 0 otherwise.



A crop group is represented by a dummy variable, indicating whether farmland ofthis type exists within 1 km of the property. Also included are variables indicating thepresence of urban parks near the property. Regarding potential multicollinearityproblems, we looked at correlation values among explanatory variables and found noindication of highly collinear explanators. Table 1 presents summary statistics of thedata used to estimate the hedonic price equation (16).

4.3. Estimation results and landscape values

Table 2 presents estimates of the parameters of the hedonic price equation (16). Thea-estimates (of the coefficients of the dwelling-specific characteristics) are mostly sig-nificant and with the expected signs. A villa commands a high premium: an averageincrease in the property value of 65% relative to an apartment or a house (the differ-ence between the latter two was found to be insignificant). As expected, closeness to acommercial area decreases the property value, and the driving time to a train stationhas a negative but insignificant effect (possibly due to non-linearities, as we expect thatliving too close or too far from a train station are both not desirable, whereas inbetween might be more desirable).

The b-estimates of vineyards, orchards and meadows are positive, as expected, andsignificant for vineyards and meadows. As livestock also entails undesirable features(noise, odour, dust), its b-coefficient is expected to be negative, as verified by the esti-mate. The c-estimates (interaction effects) are statistically insignificant.

The data and estimates in Table 2 are used to calculate DPki, specified in equation(17), for each household in the sample. We then calculate the sample average of rDPki

Table 2

Parameter estimates of the translog equation (16)

(estimates of the municipal coefficients are not presented)

Variable Coefficient SE t-statistic Prob.

Intercept 53.767 12.562 4.280 0.000

Dwelling area* 0.266 0.098 2.717 0.008Garden area* 0.027 0.012 2.327 0.022Number of bedrooms* 0.168 0.097 1.731 0.087

Number of bathrooms* 0.246 0.083 2.975 0.004Level of maintenance 0.097 0.052 1.868 0.065Age of house* �5.662 1.663 �3.404 0.001

Villa† 0.653 0.131 4.979 0.000Commercial area nearby† �0.171 0.078 �2.206 0.030Driving time to train station* �0.080 0.066 �1.219 0.226Cattle operation nearby† �0.194 0.096 �2.029 0.045

Orchards nearby† 0.316 0.280 1.126 0.263Vineyards nearby† 0.157 0.065 2.400 0.018Meadows nearby† 0.182 0.065 2.815 0.006

Orchards 9 Meadows† �0.147 0.133 �1.104 0.272Orchards 9 Vineyards† �0.057 0.266 �0.215 0.830Urban parks† 0.062 0.052 1.188 0.238

R-squared 0.863

Notes: *Continuous variable; †Dummy (0–1) variable.



to obtain the WTP for preserving crop k’s landscape by an average household in fivemunicipalities. By multiplying these average per-household WTPs by the number ofhouseholds in each municipality, we obtain estimates of WTPM

k , specified in equation(18). Dividing the latter estimates by the crop areas (see Table 3) gives the per-hectareWTP, specified in (4), and further dividing by the minimal water requirements givesthe WTP per cubic metre, specified in (13). These estimates are presented in Table 4for the two irrigation-dependent crops – vineyards and orchards. The wk values (theWTP per m3) are obtained by dividing the per-hectare values by the minimal waterrequirement [see equation (13)]. The values in parentheses use the minimal irrigationwater needs in southern Italy, where the more arid climate entails larger irrigationwater requirements.6

Table 3

Relevant data for five municipalities

MunicipalityPopulation(households)

Households nearvineyards

(%)

Households nearorchards

(%)

Vineyardsarea(ha)

Orchardsarea(ha)

Altavilla 13,643 50 33 362.3 31.1Arcugnano 3,032 60 40 80.1 15.7

Longare 2,165 100 22 239.8 17.6Sovizzo 2,641 83 42 123.0 10.4Vicenza 44,442 23 19 181.1 58.7

Table 4

Willingness-to-pay (WTP) per hectare and WTP per m3 for vineyards and orchards in fivemunicipalities.

MunicipalityWk (€/hectare)Vineyards

Wk (€/hectare)Orchards

wk (€/m3)

Vineyardswk (€/m

3)Orchards

Altavilla 14,558 16,153 16.2 (4.9) 8.1 (4.5)Arcugnano 17,574 19,499 19.5 (5.9) 9.7 (5.4)

Longare 6,983 7,748 7.8 (2.3) 3.9 (2.2)Sovizzo 13,835 15,351 15.4 (4.6) 7.7 (4.3)Vicenza 43,784 48,580 48.6 (14.6) 24.3 (13.5)

Note: The minimal water needs are 900 (3,000) m3 per hectare of vineyards and 2,000

(3,600) m3 per hectare of orchards. The values inside the parenthesis represent water needs inthe hotter climate of southern Italy. The water needs are taken from Anconelli et al. (1999).

6Minimal irrigation requirements were taken from Anconelli et al. (1999). Irrigation water in

the Vicenza region is usually charged on a per-hectare basis and rates range between € 40 and €

50 per hectare. The values in parentheses (based on water requirements in southern Italy) arefor illustrative purposes only. Obtaining genuine estimates for other regions requires estimating

the landscape values in those regions as well.



4.4. Discussion

The WTP per m3 estimates for vineyards (Table 4) range between € 7.8 per m3 inLongare and € 48.6 per m3 in Vicenza. This difference can be attributed to the differ-ent populations and vineyard areas in the two municipalities: in a more densely popu-lated municipality (Vicenza), a hectare of vineyard is likely to affect more households[cf. equations (17)], increasing the total WTP [cf. equation (18)]; indeed, this effect iswhere the public-good nature of agricultural landscape is most pronounced. Dividingthe total WTP by a smaller area (see Table 3) further increases the per-hectare valuein Vicenza compared with Longare.

The WTP per hectare of vineyard is smaller than that of orchards (Table 4). Thisdifference stems from two sources. First, the two crops affect property values differ-ently: Using equation (17), with the estimates of Table 2, reveals that the presence oforchards or vineyards near a property accounts for 10.5% or 9.5% of the propertyvalue, respectively. Second, the crop areas differ. The latter effect is likely to be smallas crop area decreases both the numerator and denominator of equation (4). The dif-ference in WTP per hectare is therefore mainly due to the first effect and is quite mod-erate. The difference in WTP per m3 (Table 4, rightmost columns) is morepronounced due to the different water needs of the crops: the minimal water inputs ofvineyards and orchards are 900 m3 per hectare (3,000 m3 per hectare in southernItaly) and 2,000 m3 per hectare (3,600 m3 per hectare in southern Italy), respectively.

5. Concluding Comments

The economic value of agricultural landscape has been found to be substantial in anumber of studies. Consequently, this environmental amenity has been used to justify(some of) the public support given to agriculture in developed economies. One way toincorporate farm landscape amenities within agricultural policy is to subsidise farmersfor the landscape amenity they provide based on their crop areas. As landscape valuesdiffer across crops, the landscape subsidies per hectare should vary accordingly. Thisapproach works well in places where land input is the limiting factor of agriculturalproduction.

In regions where irrigation water is the limiting factor of agricultural production,channelling the landscape subsidy via irrigation water pricing could be an effectivepolicy. In such cases, it is important to understand the links between crop areas andthe ensuing water demands. This allows us to infer how a certain farm policy aimed atinternalising crop landscape values affects water demands as well as how to designwater policies (pricing and quotas) to internalise the environmental amenities associ-ated with agricultural landscape. This study explains how to accomplish this task. Theempirical example demonstrates these links in our particular case.

Actual policy recommendations should be based on a comprehensive sample ofproperty prices, with spatial crop landscape characteristics. Recent development ofsatellite and GIS technologies makes such comprehensive empirical analysis feasible(see, for example, Klaiber and Phaneuf, 2010).

References

Anconelli, S., Battilani, A., Gallina, D., Genovesi, R., Guidotti, G., Mannini, P. and Pietrosi, I. Lamicroirrigazione, Regione Emilia Romagna (Bologna: C.E.R (Collana Studi e Ricerche), 1999).



Beasley, S. D., Workman, W. G. and William, N. A. ‘Estimating amenity values of urban fringefarmland: A contingent valuation approach: Note’, Growth and Change, Vol. 17, (1986)pp. 70–78.

Bergstrom, J. C. and Ready, R. C. ‘What have we learned from over 20 years of farmland ame-

nity valuation research in North America?’ Review of Agricultural Economics, Vol. 31, (2009)pp. 21–49.

Bergstrom, J., Dillman, B. L. and Stoll, J. R. ‘Public environmental amenity benefits of private

land: The case of prime agricultural land’, Southern Journal of Agricultural Economics,Vol. 17, (1985) pp. 139–149.

Bockstael, N. E. and McConnell, K. E. Environmental and Resource Valuation with Revealed

Preferences (Dordrecht, The Netherlands: Springer, 2007).Bowker, J.M. andDidychuk, D. ‘Estimation of nonmarket benefits of agricultural land retentionin Eastern Canada’,Agricultural and Resource Economics Review, Vol. 23, (1994) pp. 218–225.

Brunstad, R. J., Gaasland, I. and Vadal, E. ‘Agricultural production and the optimal level of

landscape preservation’, Land Economics, Vol. 75, (1999) pp. 538–546.Cavailh�es, J., Brossard, T., Hilal, M., Joly, D., Tourneux, F.-P., Tritz, C. and Wavresky, P.‘Pricing the homebuyer’s countryside view’, in A. Baranzini, J. Ramirez, C. Schaerer and P.

Thalmann (eds.), Hedonic Methods in Housing Markets (New York: Springer, 2008, pp. 83–100).

Drake, L. ‘The non-market value of the Swedish agricultural landscape’, European Review of

Agricultural Economics, Vol. 19, (1992) pp. 351–364.Ekeland, I., Heckman, J. J. and Nesheim, L. ‘Identification and estimation of hedonic models’,Journal of Political Economy, Vol. 112, (2004) pp. S60–S109.

Fleischer, A. and Tsur, Y. ‘Measuring the recreational value of agricultural landscape’, Euro-

pean Review of Agricultural Economics, Vol. 27, (2000) pp. 385–398.Fleischer, A. and Tsur, Y. ‘The amenity value of agricultural landscape and rural–urban landallocation’, Journal of Agricultural Economics, Vol. 60, (2009) pp. 132–153.

Hackl, F. and Pruckner, G. ‘Towards more efficient compensation programs for tourists bene-fits from agriculture in Europe’, Environmental and Resource Economics, Vol. 10, (1997)pp. 189–205.

Halstead, J. ‘Measuring the nonmarket value of Massachusetts farmland’, Journal of Northeast-ern Agricultural Economics, Vol. 13(Apr), (1984) pp. 12–19.

Hanley, N., MacMillan, D., Wright, R., Bullock, C., Simpson, I., Parsisson, D. and Crabtree,R. ‘Contingent valuation versus choice experiments: Estimating the benefits of environmen-

tally sensitive areas in Scotland’, Journal of Agricultural Economics, Vol. 49, (1998) pp. 1–15.Irwin, E. and Bockstael, N. ‘The problem of identifying land use spillovers: Measuring theeffects of open space on residential property values’, American Journal of Agricultural

Economics, Vol. 83, (2002) pp. 698–704.Klaiber, H. A. and Phaneuf, D. J. ‘Valuing open space in a residential sorting model of the twincities’, Journal of Environmental Economics and Management, Vol. 60, (2010) pp. 57–77.

Lopez, R., Shah, F. and Altobello, M. ‘Amenity benefits and the optimal allocation of land’,Land Economics, Vol. 70, (1994) pp. 53–62.

McConnell, K. ‘Optimal quantity of land in agriculture’, Journal of Agricultural and ResourceEconomics, Vol. 18(Oct), (1989) pp. 63–72.

Morey, E., Thiene, M., Desalvo, M. and Signorello, G. ‘Using attitudinal data to identify andcharacterize groups that vary significantly in their preferences for landscape preservation:A latent-class model’, Ecological Economics, Vol. 68, (2008) pp. 536–546.

OECD. Multifunctionality: Towards an Analytical Framework (Paris: OECD, 2000).Palmquist, R. B. ‘Property value models’, in K.-G. Maler and J. Vincent (eds.), Handbook ofEnvironmental Economics (North Holland: Amsterdam, 2005, pp. 763–819).

Peterson, J., Boisvert, R. and de Gorter, H. ‘Environmental policies for a multifunctional agri-cultural sector in open economies’, European Review of Agricultural Economics, Vol. 29,(2002) pp. 423–443.



Ready, R. C., Berger, M. C. and Blomquist, G. C. ‘Measuring amenity benefits from farmland:Hedonic pricing vs. contingent valuation’, Growth and Change, Vol. 28, (1997) pp. 438–458.

Rosen, S. ‘Hedonic prices and implicit markets: Product differentiation in pure competition’,The Journal of Political Economy, Vol. 82, (1974) pp. 34–55.

Scarpa, R., Campbell, D. and Hutchinson, W. G. ‘Benefit estimates for landscape improve-ments: Sequential Bayesian design and respondents rationality in a choice experiment study’,Land Economics, Vol. 83, (2007) pp. 617–634.

Schl€apfer, F. and Hanley, N. ‘Do local landscape patterns affect the demand for landscape ame-nities protection?’ Journal of Agricultural Economics, Vol. 54, (2003) pp. 21–35.

Tsur, Y. ‘On the economics of water allocation and pricing’, Annual Review of Resource

Economics, Vol. 1, (2009) pp. 513–536.



agricultural landscape value and irrigation water policy

Documents